New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
in right hands all CPU are stable, without any system administration skills none of them are stable, for server based is recommended Xeon processors.
I have a need to distribute work among many servers since a single one cannot deliver everything.
I think that's not how one should evaluate them. A better way would be to focus on what can be considered as the acceptable latency, throughput and error rate for a given system. Some other qualities can be taken into consideration too. "old-ish" or "new-ish" don't necessarily count.
Regarding upgrades, it could be as simple as switching to a newer model offered by a provider.
I don't necessarily downplay the words, but what is considered as the server grade changes over time. As said earlier, focusing on what can be considered as the acceptable latency, throughput and error rate for a given system with other qualities is a better way to evaluate hardware.
I didn't downplay Ryzen. But what I said was that two old servers would outperform a single Ryzen 5 3600 server in several ways. For example, one cannot depend on a single server if a deployed website should be accessible during system upgrades. There are other considerations too. For example, one may get two of 1 Gbps network connections instead of one. Total disc I/O performance of two parallely running old servers could be higher than that of a single Ryzen 5 3600 server. The amount of storage the providers offer with each model needs to be taken into consideration too - sometimes two old servers may provide as much as twice the storage offered with a single Ryzen 5 3600 for comparable prices. So there are many ways that two old ones will outperform a single new one.
For a certain amount of acceptable latency, for a certain amount of throughput needed, for a certain amount of acceptable error rate, for certain levels of other qualities needed, why would someone want to compare old-ish/new-ish or desktop/server grade? I would rather compare those specific parameters instead and then arrive at whether I need an old-ish/new-ish and/or desktop/server grade. For that matter, neither I prefer old systems nor I dislike new systems. Similarly, neither I prefer desktop grade nor I dislike server grade.
In short it comes down to this: My job is to figure out the components and their relationships that deliver the required functionality within the acceptable quality levels for the least amount of money.
I mean even with two ryzens it’s better price to performance then E3s or i7s from like 5 years ago. If it’s storage that’s an issue there is the AX41 which is 2x 2TB which is the same as the offerings you’re probably looking at in the auction. The only difference is the initial setup fee.
You are looking at a wrong parameter. I would instead look at price to needed performance.
Thank you all for the inputs provided! I didn't go with any of the listed models but went for an SSD vps in a bid to get the high IO performance required. Inputs you provided were considered to arrive at the current solution and will also be useful for me in future.
I came here expecting a comparison of the errata's for each CPU if you were comparing CPU stability. I don't think "stability" was the word you were looking for, probably "performance", which ended up being a storage need over CPU need as well.
I don't have data for all types, but here's some for an i7-3770 system where the cause for the instability is not known:
If you look at the above, you will see that there's a stability issue. Since not all the websites running on the system require equal level of uninterrupted operation, what I did was factoring out a part of the whole to run on an SSD VPS. It solves the stability issue to a certain degree while boosting the performance at a fractional cost.
I'll still be looking for another dedicated server to accommodate future demands that are being planned. The inputs the community provided will be useful in making a final decision.
126 restarts for a server in a year? Dude, you have defective hardware that needs to be swapped out.
You also need to do a modicum of troubleshooting, such as when does it happen, what things are running at the time, how hot things are, errors in logs, whether it reboots to apply updates, etc.
Rather than restarts, they are complete freezes. But the high uptime is due to system resets triggered by an external monitoring tool automatically.
Freezes randomly, even when the system is idling, even when temperature is normal, with no errors in logs and not due to the application of updates.
No problems have been identified by visual inspection. Thorough checks have not been done since the long downtimes are not acceptable.
Critical systems are being moved out and modified to be accessible even when this system fails. After a bit more of that, time will be available for longer downtimes and hence for thorough checking.